Ultra-wideband (UWB) is a popular technology for indoor positioning systems (IPSs) due to its robust signaling in harsh environments, through-wall propagation, and high-resolution ranging. However, the propagation of wireless signals in indoor environments is affected by nonline-of-sight (NLOS) conditions, which can lead to positively biased distance estimates. Several research works were carried out in this area, but the NLOS effect caused by the human body shadowing on the ranging performance is not properly addressed in the literature. In this paper, a commercial UWB transceiver is used to assess the impact of the human body shadowing on the ranging accuracy. A set of real-time features is combined with different machine learning algorithms for NLOS identification and mitigation. With a subset of four features, a 0.97 F1-score was obtained for NLOS identification. For NLOS mitigation, with a subset of three features, the residual error follows a Gaussian distribution, has a mean error close to zero, and an STD of 0.67 m.

中文翻译：

---

贴身式 UWB 收发器的实时 NLOS 识别和缓解功能选择


超宽带 (UWB) 是室内定位系统 (IPS) 的热门技术，因为它在恶劣环境下具有强大的信号传输能力、穿墙传播和高分辨率测距功能。然而，室内环境中无线信号的传播会受到非视距 (NLOS) 条件的影响，这可能会导致距离估计出现正偏差。在该领域开展了多项研究工作，但文献中没有适当解决人体阴影对测距性能造成的非视距效应。本文使用商用 UWB 收发器来评估人体阴影对测距精度的影响。一组实时特征与不同的机器学习算法相结合，用于 NLOS 识别和缓解。通过四个特征的子集，NLOS 识别的 F1 分数为 0.97。对于 NLOS 缓解，使用三个特征的子集，残差遵循高斯分布，平均误差接近于零，STD 为 0.67 m。